On Thu, Apr 19, 2012 at 5:14 PM, Chris Angelico <ros...@gmail.com> wrote:
> The question then is whether to choose or auto-detect. Attempting to > auto-detect could be quite inefficient; imagine if you have to call on > ssh every couple of seconds, and something in $PATH is on a slow > network share (or on a floppy drive, if you still have one of those). > Yes, it's a poor choice of executable path, but it's perfectly legal, > and the detection could get pathologically expensive. At very least, > consider caching the result. > The wrapper class for autodetecting could cache locations to speed things up. bash does this normally (hence the "rehash" command), but probably not across distinct shell processes. I'm not a big fan of hardcoding paths; they have a way of changing. I prefer $PATH, or $PATH search that caches. > That way when Haiku takes the world by storm, your porting effort should > be > > smaller. ^_^ (Or something. I really, really hope that Windows, OS/X > and > > Linux won't be the only major OS's for all time) > > Heh. Point. But the detection technique does depend on 'ssh' being the > secure shell client and accepting parameters exactly like you're > looking for - imagine if the Haiku (or whatever) ssh client takes > slightly different params. How do you handle that with autodetection? > True, that gets more complicated. For commands that look faintly similar but aren't really very similar, you'd probably have a less-generic wrapper class that gets a usage message from the ssh binary or something. The generic one is just for the simple case.
-- http://mail.python.org/mailman/listinfo/python-list
